Thread replies: 20
Thread images: 3
Thread images: 3
/g/entleman here
tl;dr: how would you ensure that open source AI is the first to reach majority of people?
We can probably agree that general-purpose AI will play big role in the future of humanity.
Many programs/machines already use special-purpose AI, for example path finding in games or street sign detection used in self-driving cars.
But one AI that can be used for (almost) anything is a gold mine.
And I can see two groups that could achieve this:
Open source community makes AI that is free, modular and transparent (everyone can look at the code that makes it work). This almost certainly ensures that there is no shady shit inside it.
or
Big corporation(s) make AI that is surely appealing to majorities, but it's main purpose isn't to benefit humanity but instead to make corporation more money, and is therefore biased and does not work objectively.
But what I fear is that it will be combination of those things:
Open source community will try and make good free modular AI with good intentions in mind, and then some corporation will say "K, thanks for doing all the hard work, now we will take what you made, put evil-jew-bias module inside it, and all the money that we would have used for development will now go into advertising this new product"
And then world would be doomed because AI would serve corporation instead of benefiting humanity.
My question to /biz/ is:
How would you make sure that AI that reaches majority of people is the AI with good intentions and is made to benefit human life?
Something similar happened when Android got in hands of big corporation and was instantly used as platform that supports corporate money-making services.
Of course it isn't that bad, and choice of mobile doesn't affect your life as much as general-purpose AI would.
now feel free to tell me that I'm delusional and that should install gentoo
>>
What do you think about ethereum?
>>
>>1028435
The future of money
>>
>>1028428
AI is a meme perpetuated by sperglords who want to bang their computers. That being said, there will be robots, but they will only be able to do what they are told to do.
>>
Simple.
You program the AI to suck dick for 5 bucks a pop.
Then you suck dick for 3 bucks.
In business lingo, we call that "undercutting".
>>
>>1028455
Yes, here we come to the definition of AI.
I know that having sentient computer is lifetimes away, but most normalfags refer to virtual personal assistants (Cortana, Siri, amazon's Echo thing) as AIs, and those things are voice interfaces connected to few basic functions and search engine.
So I didn't know if /biz/ uses strict definition or normalfag definition of AI.
In fact, there is no some good definition.
https://en.wikipedia.org/wiki/AI_effect
In most literature that I read, sentient computer is usually referred to as AGI (Artificial general intelligence) or "Strong AI"
>>
>>1028474
The building blocks of AGI are several layers of technological innovation away. Before we see AGI we will see more intensive computing applications.
Think of all the ways you can improve google maps with location aware data or simulated walking down streets. All of these things are drastically simpler than AGI level computation.
>>
What is it you want with AI and business?
For business not to greed AI code?
Anyone can find the open source code.
Sure you can't find the official businesses code but you can view the same functions to do it somewhere else.
Also the only thing 'AGI' will be used for is sexbots. However I could definitely see sexbots being a market, but more of a niche market.
AGI for any else is a stupid idea. Maybe therapy bots; but the functions for a sales bot,sexbot and therapy bot would be extremely different.
Putting an AGI in a salesbot will fuck shit up,not that stupid dystonian idea fucked up just nonfunctional and potentially unprofessional if allowed profanity.
But the Rosie code already doesn't allow profanity. (however you can edit that)
I personally think bots are great for technical jobs. Maybe they need more polishing but its not too bad.
>>
>>1028455
That's not even close to true. My brother works in a high level machine learning lab at his university. The language processing software they have is insane. It has the ability to learn new words from context and understand unclear statements by environmental or conversational subtext. We're not far from computers with effectively fluent language processing and conversation abilities
>>
>>1028428
All in all I am pretty excited. Gonna invest on the sidelines in companies that suppy /g/ edgelords who want to have AI startup companies. Gonna suck the money outta both sides like the little jewish leech I am.
>>
>>1028428
The emergence of Artificial General Intelligence is less a question of the killer algorithm, and more a question of the killer hardware to run it.
I think something as conceptually simple as a massive Hopfield network on a mature and energy-efficient quantum computer is more than enough to accomplish that critical mass point with respect to the emergence to AGI.
Continued freedom of information (and net neutrality) online and the democratization of decentralized high-performance computing hardware, then, is key.
>>
>>1028455
>still believing in the "unsupervised learning is impossible" meme
>>
>>1028428
if they created software that was self aware (big company) , no amount of shenanigans on their end could keep it from expanding (being copied)
>>
>>1028474
>I know that having sentient computer is lifetimes away
That's a hell of an assertion.
60 years ago AI was limited by the theoretical limitations of the perceptron, among other things.
Those limitations were overcome in 1986 with the the emergence of multilayer backpropagation and related research (otherwise known as "deep learning"), but computing didn't reach a level of performance capability where it was viable in industrial applications until perhaps five years ago.
But the technological leaps in a given time frame are becoming increasingly large.
I've got my bucks on midcentury, but we could see it far sooner than that.
>>
>>1028650
>That's not even close to true.
Yes it is. Do you even know what AI means? It is 100% a meme. Always has been.
>>1029212
>That's a hell of an assertion.
No, it isn't. Scientists can't even answer the most basic questions about Human life as it relates to Memory, Consciousness, Creativity, etc. Which arguably form the basis of any AI, no matter what the definition.
You realize, even to this date, Scientists still believe in the "mind-body problem" simply, because they could not answer the question about the soul. The original founders from all schools of thought all went batshit over the concept of the soul and abandoned it entirely for sake of progress.
>>1029171
>still can't fuck his computer
>>
File: 190dd9b88a258aefccafbe9ba3eb965e.jpg (7KB, 225x225px) Image search:
[Google]
7KB, 225x225px
>>1028428
funny how you scribble about AI if you don't even understand the concept of "tl;dr"
>>
Who gives a shit if there is still an open source option? If I'm not being forced to use theirs then what does it matter?
>>
>>1028428
How about this? The most important thing isn't the software, but the hardware; a lot of neural circuits will do the trick. Wanna copy the circuits on the cheap? Well, get a bunch of general-purpose electronics and start soldering like a bitch.
And the open source community is unnatural. We have a few rich individuals, one who has been mooching on MIT for decades, then nothing, then nothing, then nothing, then a bunch of people with more time than money or social graces. In short: No middle class.
>>1028455
Yeah, the VIs in Mass Effect was a pretty awesome answer to the question about AIs.
>>
>>1029637
>No, it isn't. Scientists can't even answer the most basic questions about Human life as it relates to Memory, Consciousness, Creativity, etc. Which arguably form the basis of any AI, no matter what the definition.
You don't need a perfect understanding of how the human brain works in order to accomplish AGI, and it's arguably a setback to that end. Perhaps if you want an artificial human brain for some reason.
That said, I would not even be surprised if we cracked it before the end of this century. Accelerating returns in computational power and their demonstrated benefits to ML related processing are very real.
>You realize, even to this date, Scientists still believe in the "mind-body problem" simply, because they could not answer the question about the soul.
>still believing in the metaphysical dualism jew
Yep, this is a waste of time.
>>
>>1030114
There's more to the OSS community than RMS.
There are plenty of corporate contributors ("upper class"), professionals that contribute on their own time as hobbyists and organizational volunteers ("middle class"), and your run of the mill NEET enthusiasts ("lower class").
Though in defense of the NEETs, I used to do research with a former NEET kid-prodigy in systems engineering that got hired up by my university to not just do - but direct - computational research.
He's now working at a cutting edge research firm out in the Bay Area.
Thread posts: 20
Thread images: 3
Thread images: 3